据报道，Meta近日揭晓了其两个全新的24k GPU集群，旨在进一步推进其在人工智能领域的研究与开发。这些集群将用于训练Llama 3，一种大语言模型。这些新集群基于Meta的AI Research SuperCluster（RSC），该集群于2022年首次亮相。新集群的设计、性能和软件的详细信息在Meta的博文中得到了分享。
与之前包含16,000个Nvidia A100 GPU的旧集群相比，新集群的GPU数量大幅增加至24,576个Nvidia Tensor Core H100 GPU。这一显著增长使得新集群能够支持更大、更复杂的模型，为生成式AI产品的研发铺平了道路。
Meta表示，这些新集群将在自然语言处理、语音识别和图像生成等领域发挥关键作用。随着人工智能技术的不断进步，这些领域预计将迎来更多创新。
英文标题：Meta Unveils AI Powerhouses
英文关键词：Meta AI, GPU Clusters, Llama 3
英文新闻内容:
Meta recently unveiled its new 24k GPU clusters, designed to further advance the company’s research and development in the field of artificial intelligence. These clusters will be used to train Llama 3, a large language model. Meta shared the design, performance, and software details of the new clusters in a recent blog post.
Compared to the previous cluster, which contained 16,000 Nvidia A100 GPUs, the new clusters feature a significant increase to 24,576 Nvidia Tensor Core H100 GPUs. This growth allows the new clusters to support larger and more complex models, paving the way for advancements in generative AI product development.
Meta stated that the new clusters will play a crucial role in areas such as natural language processing, speech recognition, and image generation. As AI technology continues to evolve, these areas are expected to see more innovation.
【来源】https://www.datacenterdynamics.com/en/news/meta-reveals-details-of-two-new-24k-gpu-ai-clusters/
Views: 1